Because You Watched

You share the account — most families do now, one login for the streaming, the music, the shopping, split among everyone to save the cost — and so when they die, their taste does not leave with them. It stays in the account, baked into the recommendations, and the algorithm goes on serving it. Because you watched, the row says, and names something they watched, and offers you more of it. Recommended for you, it says, and the you is partly them. The feed keeps suggesting the kind of thing they liked, keeps steering toward their preferences, keeps building out the portrait of a taste it has no idea has stopped generating new data — because to the algorithm, the account is a single continuous person, and that person’s history includes everything the dead one ever watched and wanted, and so the machine goes on faithfully chasing the taste of someone who is gone, recommending to the living the things it learned from the dead.

There is a low strangeness to it that takes a while to surface, because at first you don’t notice — the recommendations are just the recommendations, the ambient noise of a feed. And then one day a row appears, because you watched, and it names a show you never watched, a whole genre you have no interest in, and you realize with a small jolt that the you the machine means is not you at all, it is them — that the feed has been quietly half-shaped by a dead person’s preferences this whole time, that some fraction of everything it offers you is really an offering to someone who cannot accept it, that you have been sharing your recommendations with a ghost. The algorithm does not distinguish. It cannot see the seam where one person stopped and another kept going. It has merged you, blended two lifetimes of taste into a single profile, and it serves the blend back without knowing that half of it is a memorial.

The taste that outlives the person

We do not usually think of taste as a thing that can be left behind, but it is, and the shared account is where you see it — because taste turns out to be one of the most persistent traces a person leaves, more durable in some ways than the accounts themselves. What a person liked, the specific texture of their preferences, the particular things that drew them, the pattern of their attention across a thousand small choices — the algorithm captured all of it, not as a list but as a shape, a gravitational pull that goes on bending the feed toward the kinds of things they’d have wanted. And that shape persists. Long after the last thing they watched, the recommendations still lean the way their taste leaned, still surface the things they’d have surfaced, still carry the fingerprint of a sensibility that has otherwise left the world. The machine is, without meaning to be, the last faithful keeper of their taste — still asking, in effect, what would they like? and answering, and offering it up, month after month, to a household that mostly no longer remembers the feed is doing it.

And there is something in that which is almost tender, if you turn it the right way. Because everything else about them is being sorted, slowly, into the past — the accounts closed, the things given away, the presence receding. But the algorithm never got the news, and so in the one place, the feed, they are still a current taste, still an active set of preferences, still someone the machine is trying to please, still being asked what do you want to watch next? as though they might yet sit down and choose. The recommendation engine treats them as a living appetite, ongoing, unfinished — and there are worse things to be, in the flat memory of a machine, than a taste still being served, a person the algorithm is still, faithfully, trying to delight.

The self you can’t unblend

The part that stays with me is that you cannot separate it out anymore. Over months of a shared account, your choices and theirs braided together into one profile, and the algorithm learned you both as a single person, and there is no untangling which recommendations descend from your history and which from theirs — the feed is a genuine blend now, a merged taste, two sensibilities folded into one machine-portrait of a person who is really two people, one living and one gone. You could try to purge it, of course — clear the history, remove their profile, tell the machine to forget what it learned from them, scrub the account down to only your own taste. But to do that is to reach in and delete the last active version of their preferences, to un-blend the self, to tell the algorithm to stop asking what would they like — and a lot of people find they don’t want to. They leave the histories braided. They let because you watched keep naming the things the dead one watched. They let the feed go on being a shared feed, half theirs, because the blend is the last place the two of you are still, in some small machine-deep way, one household with one taste, choosing together, even now.

I am not going to tell you whether to keep the profiles merged or clear them, because it is a small thing and a genuinely personal one, and there is no wisdom to dispense about a streaming account. I only wanted to notice what the shared feed quietly becomes — a place where a dead person’s taste outlives them and goes on being served; where the algorithm, blind as ever, keeps them current, keeps asking what they’d want, keeps folding their preferences into the household’s; and where, if you leave it, your recommendations stay braided with theirs indefinitely, two selves blurred into one by a machine that never learned to tell you apart, still saying because you watched, still meaning, without knowing it, because you both did — which is, of all the ways to be remembered, one of the stranger and quieter and more oddly companionable ones: kept alive as a taste, in a feed, still choosing alongside the people you left.